House mice live freely all over the small Welsh island of Skokholm The population has been in existence for over 70 years and was studied intensively from 1960 to 1969. A presumed random sample from the island mice of 59 males and 58 females was killed, in September and data were collected on size, organ weights and structure, chemical composition, haematological traits, reproductive condition, and isozyme variants. The animals were allotted to eight age-classes on the basis of tooth wear, such that the oldest group were over a year old and had survived the previous winter, while the others had all been born in the current breeding season. Isozyme genotype frequencies showed that natural selection was operating on the mice, so that the age groups were not genetically homogeneous. Moreover discriminant function analysis produced good distinctions between isozyme phenotypes. This meant that any differences between the age-classes had to be determined on the data adjusted for genotypes as well as size. Nevertheless, a discriminant analysis could still recognize the different age-categories. Multiple regression studies made the identification of the contribution of relevant factors possible. A number of characters were shown to be age-correlated, and their importance varied in mice from different habitats; no one trait could adequately describe the observed changes with age. We conclude that ageing studies carried out on ‘standardized’ laboratory animals in a necessarily over-simplified experimental environment may give results misleading to gerontologists.